This study aims to examine how health behaviors and lifestyles evolve during old age in the United States (US) and Japan. Although individual health behaviors - including alcohol consumption, smoking, physical activity, and weight management - have been widely studied, we currently know little about how these behaviors change in later life. Moreover, few studies have moved beyond the analysis of individual health behaviors to examine configurations of multi-behavior profiles within individuals (i.e., health lifestyles). The proposed research has the following specific aims: (1) To describe how individual health behaviors change during old age;(2) To describe the prevalence of various health lifestyles during old age. We define a health lifestyle as a multidimensional profile representing one's status with respect to various independent health behaviors measured simultaneously;(3) To assess the extent to which interindividual variations in late life health behavior trajectories and lifestyle profiles are patterned by key social status indicators (e.g., ethnicity, socioeconomic status, and gender) and changing psychosocial factors (e.g., personal control beliefs, social resources, and stress);(4) To analyze the interrelationships between health outcomes and various health behavior trajectories and lifestyle profiles during late life;(5) To compare and contrast the prevalence, determinants, and health impact of various late life health behavior trajectories and lifestyle profiles in the US and Japan. Data for the proposed project come from four nationwide panel surveys conducted in the United States: the Health and Retirement Study (HRS, ICPSR 6854), Well-Being Among the Aged: Personal Control and Self-Esteem (WBA, R01 AG009221), the Americans'Changing Lives study (ACL, ICPSR 3394), and Aging, Status, and Sense of Control (ASOC, ICPSR 3334);and one nationwide panel survey of Japanese elderly: the National Survey of Japanese Elderly (NSJE, ICPSR 6842). These data sets are ideal for this research because each includes repeated observations, over multiple-year periods, of a wide variety of health behaviors, and together they allow for multiple comparisons and the cross-validation of our research findings between different cultures. These study aims will be addressed primarily with hierarchical linear modeling, cluster analyses, group-based trajectory modeling, and multinomial logit analyses. PUBLIC HEALTH RELEVANCE: Tracking age-related changes in health behaviors may reveal new information about how the dynamic facets of our own behaviors influence health in old age, and could help us in targeting interventions to those elderly who are particularly vulnerable to patterns of worsening health behaviors. Furthermore, examining profiles of multiple behaviors simultaneously (i.e., lifestyles) allows for the examination of the joint effects of multiple behaviors on health, perhaps revealing the value of tailoring interventions to specific lifestyles, rather than specific behaviors. Finally, additional insights into the health behaviors and lifestyles of US elderly can be gained through comparisons with elderly from other highly developed, yet culturally distinct, countries such as Japan.